Lead Investigators:

Yeung Lam, Eric Yuen, Eric Graham, William Kaiser (UCLA)

Overview

The objective of this project is to modify the existing NIMSRD terrestrial system to suit the requirements of a semi-permanent installation at a NEON field site. Whereas the existing system has been targeted towards rapid deployments at remote sites for up to a weeks worth of data collection, the new system will be targeted towards long term operation with an emphasis on reliability.

Also, this project includes development of the NIMS RD TCP/IP Server/Client model is to develop a framework in which applications can be developed to control the NIMS RD system. The server application should effectively provide a user or application with a set of commands that provide complete control of the NIMS RD node. Additionally, the server should assure a level of reliability that will allow a client application to assume consistent behavior even in the case of faults, such as errors returned by the motor controller.

Approach

Infrastructure:
The current housing for the motors, controllers, and NIMSRD computer will be replaced by a two part enclosure. A weatherproof, but adequately ventilated structure will house the NIMSRD computer and controllers and a separate mounting fixture will be created for the motors. The controller housing will be modular such that the basic enclosure can be upgraded with additional infrastructure to accommodate different environmental requirements.

Sensor node will be made weather resistant. This will include sealed bearings on the shuttle and a cable brush to remove foreign matter that may collect on the cable.

Once the new infrastructure has been developed, maintenance schedules and procedures will be developed to ensure proper functionality of the system over an extended period of time.

A networked UPS will be placed between the NIMSRD system and its power source. This will prevent system outages during brief power outages that may occur. Additionally, should the system ever require a power cycle, this will facilitate that remotely.

One or several new localization methods will be added including possibly a combination of the following: hall effects sensors, optical encoder, physical limit switch, differential GPS.

The current motors will be replaced with motors which have a gear ratio more appropriate to the application. The new motors will retain the same control interface and be operated using the same motor drivers.

Software:
The current software used to facilitate control and raster operations of the system will be replaced by a multithreaded socket based server daemon. The new software will provide a standard socket interface to client applications for control as well as data acquisition. The new system will support end to end sensor to sensorbase functionality and provide real time web accessible data streaming.

A layer of system and network monitoring will be put in place to continually ensure proper functionality.  System health parameters such as enclosure temperature will be uploaded to the sensorbase repository. Email alerts for suspect values will be employed to alert users of possible malfunction. Periodic check on network connectivity from a remote system will alert users if network connectivity is down.

Server Daemon:
The NIMS RD TCP/IP server provides a standard set of commands that allow a user or client application to control the NIMSRD node and retrieve positional and system health data from the node. The daemon accepts two simultaneous TCP/IP connections: one connection has administrative permissions and the other has observer permissions. The administrative permissions allow the user to adjust acceleration and velocity settings, move the node, stop movement of the node, get the position of the node, and retrieve the system status of the node. The observer permissions only include getting the position of the node and retrieving the system status of the node.

The server daemon provides a framework for developing client applications to drive the node for various sampling tasks. These client applications could include, but would not be limited to: simple raster scripts with defined sampling positions and dwell times, real time adaptive sampling algorithms that determine the sampling strategy of the node based on sensor readings, and a real time control interface to facilitate manual control of the system. Additionally, by using TCP/IP, client applications can be developed and executed remotely and on any platform.

To address the issue of system reliability, confirmation messages are sent to the user or client application for every command that attempts to change system settings or move the system. In the situation that the underlying motor driver returns an error, the server will analyze the error and inform the client or user of the error and whether the move could be confirmed as successful or not. In this way, the client application can be designed to explicitly handle known faults and reach in a consistent way.

Raster Client:
The raster client that was developed for the terrestrial experiments at the James Reserve leverages the TCP/IP interface provided by the server daemon in order to perform a raster scan as well as fuse real time sensor data with positional data. The client allows the user to adjust the sampling interval and dwell times as well as define sensor parameters such as sensor equilibrium times for a given experiment.

Data is recorded both in raw format at the highest possible sampling frequency as well as in a format defined by the user for the particular experiment. Both data sets are uploaded in real time to sensorbase.org as well as being recorded locally in the case where no internet connection is available. By providing the raw data set and the refined data set, the system is able to provide a data set that the user can apply immediately to the designed experiment while also providing sufficient data to be used for further analysis.

The NIMS RD system at JR sends its data to the CENS database, SensorBase.org automatically.  Below is an image of the SensorBase interface where the NIMS RD at the James Reserve data is available.  Users can select from the data columns available and download the data in XML or comma delimited ASCII.

Visualization tools are the next step towards making this data useable.  For example, during a 48-hour test run this Spring, data was collected from the NIMS RD system running at the James Reserve every 20 seconds and was sent to SensorBase.   A webpage was developed to view this data as it was collected in real time along with historical data (see screen capture below).

The data collected included air temperature, PAR, relative humidity, and net radiation from the mobile NIMS RD shuttle as well as soil sensors on a stationary base station.  System data from the NIMS node was also collected, including motor temperature and target positions. 

The list of sensors was displayed with their last recorded value. Selecting one of the sensors from the list displayed a graph of its values over a specified time period.  A live streaming camera pointed at the system from a tower camera was used to verify the movements of the system.

Future Directions

The current NIMSRD terrestrial system uses a Campbell datalogger mounted on the mobile node to interact with sensors. This data is sent via a serial connection to the NIMSRD computer and the data is time stamped and merged with positional data from the NIMSRD system. A raster program on the NIMSRD control computer operates independently and gets no feedback from the sensors. This scheme provides little more than time stamps to coordinate positional and sensor data.

It is our objective not only to replace the Campbell datalogger with the compactRio, but to shift control of the system over to the compactRio as well. In this way, the NIMSRD control computer will merely be another sensor device connected to the compactRio.  The CompactRIO will contain the raster scripts and the NIMSRD server daemon will provide a serial interface similar to that of other serial based sensors for configuration, control, and data acquisition. The same blue heat/copperlink interface that currently connects the Campbell datalogger to the NIMSRD computer will be used to connect with the compactRio.

Initially, the NIMSRD system will make use of sensorbase.org as a repository for real time data. Data from the sensors (including positional data) will be uploaded real time to the sensorbase.org repository.  Client software will be written to manage projects. Project management would include creation and configuration of raster scans and configuration of sensors. Upon integration of the compactRio this can be expanded to include NEON specific data repositories.

Development Plan

February JR Deployment
Software Development:
Multithreaded socket based server daemon and raster client
System health monitoring and reporting
Network monitoring and reporting
Sensor to sensorbase end to end data system

March JR Deployment
Weatherproof housing for motor controller, NIMSRD computer
Separate weatherproof housing/mounting fixture for motors
Localization via combination of dgps, optical encoder, hall effects sensors, or physical limit switch.
Networked UPS
Project management via sensorbase.org

April JR Deployment
Serial NIMSRD interface for compactRio integration
Defined maintenance schedule and procedures
Designs for weatherproof modules for different environments

Deployment Objectives
James Reserve:
February deploytment:
1-2  transects running for 24 hours perpendicular to AMARS transect.
Test out new NIMSRD server daemon software.
March deployment:
Test out new infrastructure.
Test out localization.
April deployment:
CompactRio integration

La Selva
Tropical rainforest permanent installation with sensor to sensorbase end to end system.

Los Amigos
Remote tropical rainforest permanent installation with v-sat uplink and sensor to sensorbase end to end system

Stunt Ranch
NIMSRD system similar to that of JR.